The recording of surgical procedures for and diagnoses of nonmalignant gynecological conditions in the Danish National Hospital Registry in 1977-1988 were evaluated by comparison with discharge summaries for a sample of 4,919 women. A serious problem was found in the validity of the diagnoses. Little variation was seen over the period of observation. The coding of operations was considered valid. The results of the study raise concern in view of the widespread use of the registry data in medical research. The authors suggest that data from the registry should be thoroughly evaluated before firm conclusions based on such data are published, and that efforts to improve the validity of the data in the Danish National Hospital Registry are increased.